yt_dlp/extractor/medaltv.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_str
   5 from ..utils import (
   6     ExtractorError,
   7     float_or_none,
   8     format_field,
   9     int_or_none,
  10     str_or_none,
  11     traverse_obj,
  12 )
  13
  14
  15 class MedalTVIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?medal\.tv/games/[^/?#&]+/clips/(?P<id>[^/?#&]+)'
  17     _TESTS = [{
  18         'url': 'https://medal.tv/games/valorant/clips/jTBFnLKdLy15K',
  19         'md5': '03e4911fdcf7fce563090705c2e79267',
  20         'info_dict': {
  21             'id': 'jTBFnLKdLy15K',
  22             'ext': 'mp4',
  23             'title': "Mornu's clutch",
  24             'description': '',
  25             'uploader': 'Aciel',
  26             'timestamp': 1651628243,
  27             'upload_date': '20220504',
  28             'uploader_id': '19335460',
  29             'uploader_url': 'https://medal.tv/users/19335460',
  30             'comment_count': int,
  31             'view_count': int,
  32             'like_count': int,
  33             'duration': 13,
  34         }
  35     }, {
  36         'url': 'https://medal.tv/games/cod-cold-war/clips/2mA60jWAGQCBH',
  37         'md5': 'fc7a3e4552ae8993c1c4006db46be447',
  38         'info_dict': {
  39             'id': '2mA60jWAGQCBH',
  40             'ext': 'mp4',
  41             'title': 'Quad Cold',
  42             'description': 'Medal,https://medal.tv/desktop/',
  43             'uploader': 'MowgliSB',
  44             'timestamp': 1603165266,
  45             'upload_date': '20201020',
  46             'uploader_id': '10619174',
  47             'thumbnail': 'https://cdn.medal.tv/10619174/thumbnail-34934644-720p.jpg?t=1080p&c=202042&missing',
  48             'uploader_url': 'https://medal.tv/users/10619174',
  49             'comment_count': int,
  50             'view_count': int,
  51             'like_count': int,
  52             'duration': 23,
  53         }
  54     }, {
  55         'url': 'https://medal.tv/games/cod-cold-war/clips/2um24TWdty0NA',
  56         'md5': 'b6dc76b78195fff0b4f8bf4a33ec2148',
  57         'info_dict': {
  58             'id': '2um24TWdty0NA',
  59             'ext': 'mp4',
  60             'title': 'u tk me i tk u bigger',
  61             'description': 'Medal,https://medal.tv/desktop/',
  62             'uploader': 'Mimicc',
  63             'timestamp': 1605580939,
  64             'upload_date': '20201117',
  65             'uploader_id': '5156321',
  66             'thumbnail': 'https://cdn.medal.tv/5156321/thumbnail-36787208-360p.jpg?t=1080p&c=202046&missing',
  67             'uploader_url': 'https://medal.tv/users/5156321',
  68             'comment_count': int,
  69             'view_count': int,
  70             'like_count': int,
  71             'duration': 9,
  72         }
  73     }, {
  74         'url': 'https://medal.tv/games/valorant/clips/37rMeFpryCC-9',
  75         'only_matching': True,
  76     }, {
  77         'url': 'https://medal.tv/games/valorant/clips/2WRj40tpY_EU9',
  78         'only_matching': True,
  79     }]
  80
  81     def _real_extract(self, url):
  82         video_id = self._match_id(url)
  83
  84         webpage = self._download_webpage(url, video_id, query={'mobilebypass': 'true'})
  85
  86         hydration_data = self._search_json(
  87             r'<script[^>]*>[^<]*\bhydrationData\s*=', webpage,
  88             'next data', video_id, end_pattern='</script>', fatal=False)
  89
  90         clip = traverse_obj(hydration_data, ('clips', ...), get_all=False)
  91         if not clip:
  92             raise ExtractorError(
  93                 'Could not find video information.', video_id=video_id)
  94
  95         title = clip['contentTitle']
  96
  97         source_width = int_or_none(clip.get('sourceWidth'))
  98         source_height = int_or_none(clip.get('sourceHeight'))
  99
 100         aspect_ratio = source_width / source_height if source_width and source_height else 16 / 9
 101
 102         def add_item(container, item_url, height, id_key='format_id', item_id=None):
 103             item_id = item_id or '%dp' % height
 104             if item_id not in item_url:
 105                 return
 106             width = int(round(aspect_ratio * height))
 107             container.append({
 108                 'url': item_url,
 109                 id_key: item_id,
 110                 'width': width,
 111                 'height': height
 112             })
 113
 114         formats = []
 115         thumbnails = []
 116         for k, v in clip.items():
 117             if not (v and isinstance(v, compat_str)):
 118                 continue
 119             mobj = re.match(r'(contentUrl|thumbnail)(?:(\d+)p)?$', k)
 120             if not mobj:
 121                 continue
 122             prefix = mobj.group(1)
 123             height = int_or_none(mobj.group(2))
 124             if prefix == 'contentUrl':
 125                 add_item(
 126                     formats, v, height or source_height,
 127                     item_id=None if height else 'source')
 128             elif prefix == 'thumbnail':
 129                 add_item(thumbnails, v, height, 'id')
 130
 131         error = clip.get('error')
 132         if not formats and error:
 133             if error == 404:
 134                 self.raise_no_formats(
 135                     'That clip does not exist.',
 136                     expected=True, video_id=video_id)
 137             else:
 138                 self.raise_no_formats(
 139                     'An unknown error occurred ({0}).'.format(error),
 140                     video_id=video_id)
 141
 142         # Necessary because the id of the author is not known in advance.
 143         # Won't raise an issue if no profile can be found as this is optional.
 144         author = traverse_obj(hydration_data, ('profiles', ...), get_all=False) or {}
 145         author_id = str_or_none(author.get('userId'))
 146         author_url = format_field(author_id, None, 'https://medal.tv/users/%s')
 147
 148         return {
 149             'id': video_id,
 150             'title': title,
 151             'formats': formats,
 152             'thumbnails': thumbnails,
 153             'description': clip.get('contentDescription'),
 154             'uploader': author.get('displayName'),
 155             'timestamp': float_or_none(clip.get('created'), 1000),
 156             'uploader_id': author_id,
 157             'uploader_url': author_url,
 158             'duration': int_or_none(clip.get('videoLengthSeconds')),
 159             'view_count': int_or_none(clip.get('views')),
 160             'like_count': int_or_none(clip.get('likes')),
 161             'comment_count': int_or_none(clip.get('comments')),
 162         }